Clive Ebomagune Asuai; Gabriel Ogbogbo; Houssem Hosni; Muhammad Ibrahim Khan

The Chromatic Gradient Anomaly Network (CrGAN): Exploiting Second-Order Spatiotemporal Inconsistencies for Deepfake Video Detection

PDF (1894KB), PP.139-164

Views: 0 Downloads: 0

Author(s)

Clive Ebomagune Asuai ^1,* Gabriel Ogbogbo ² Houssem Hosni ³ Muhammad Ibrahim Khan ⁴

1. Department of Cyber Security, Delta State Polytechnic, Otefe-Oghara, 333106, Nigeria

2. Department of Statistical Sciences, Delta State Polytechnic, Otefe-Oghara, 333106, Nigeria

3. Department of Computer Engineering, Université de La Rochelle, 17000 La Rochelle, France

4. Department of Electrical and Computer Engineering, University of Illinois, Urbana-Champaign, USA

* Corresponding author.

DOI: https://doi.org/10.5815/ijwmt.2026.02.10

Received: 23 Jan. 2026 / Revised: 14 Feb. 2026 / Accepted: 14 Mar. 2026 / Published: 8 Apr. 2026

Index Terms

Deepfake Detection, Video Forensics, Chromatic Gradient Anomaly Network, Spatiotemporal Inconsistencies, Second-Order Artifacts, Anomaly Localization, Generative Model Artifacts, Digital Media Integrity

Abstract

Unregulated accessibility to the latest deepfake technologies presents escalating, unprecedented threats to personal security, public trust, and democratic integrity, owing to the ever-increasing sophistication and realism of these forgeries. The biggest challenge is the inability of human verification to ascertain the original from the forgeries. Therefore, this research aims to establish an initial framework of detection and verification. This research presents a completely new way of detecting manipulation by looking for second-order spatiotemporal inconsistencies in chromatic energy distributions, as opposed to existing deepfake detection methods that rely on complicated multi-stream architectures or first-order pixel-level features. The theoretical importance comes from the fact that generative models can convincingly copy static visual features, but they always fail to keep colour and texture changes that make sense in both space and time. The Chromatic Gradient Anomaly Network (CrGAN) is an architecture that will be built and tested to capture changes of the various components of a video over time in order to reveal patterns of inconsistency between the spatiotemporal levels of a video and the changes of its chromatic components. This method is useful in two ways: first, it gets state-of-the-art detection accuracy without needing complicated multi-modal fusion; second, and more importantly, it lets forensic analysts see exactly where and how a video was changed at the pixel level, which is very important for legal and investigative purposes. One of the most important contributions of this research is the analysis of the second-order derivatives (in this case, the Chromatic Gradient Fields) of the Spatiotemporal Chromatic Energy Distributions, revealing the synthesis boundary of temporally sparse flickers and the physically implausible discontinuities of the blend. The results for CrGAN demonstrate the highest level of diagnostic confidence, reporting a detection rate of 97.9%, and most importantly a level of pixel-wise localized mapping of the detected region that is statistically differentiated from other detection models, achieving state-of-the-art performance while maintaining architectural simplicity. This is a big change in how deepfake detection works: it moves complexity from model architecture to forensic signal representation, which makes the solution more elegant, easier to understand, and more generalizable.
In conclusion, this study validates how targeting second-order spatiotemporal inconsistencies using chromatic gradients not only acts as an efficient detection mechanism but also as an interpretable tool in the combat against digital deception by identifying the how and where of video forgery.

Cite This Paper

Clive Ebomagune Asuai, Gabriel Ogbogbo, Houssem Hosni, Muhammad Ibrahim Khan, "The Chromatic Gradient Anomaly Network (CrGAN): Exploiting Second-Order Spatiotemporal Inconsistencies for Deepfake Video Detection", International Journal of Wireless and Microwave Technologies(IJWMT), Vol.16, No.2, pp. 139-164, 2026. DOI:10.5815/ijwmt.2026.02.10

Reference

[1]Clive, O. K. Nana, and I. E. Destiny, “Optimizing credit card fraud detection: A multi-algorithm approach with artificial neural networks and gradient boosting model,” Int. Res. J. Mod. Eng. Technol. Sci., vol. 6, no. 12, pp. 2582–5208, 2024.
[2]Clive, C. T. Atumah, and A. Agajere Joseph-Brown, “An improved framework for predictive maintenance in Industry 4.0 and 5.0 using synthetic IoT sensor data and boosting regressor for oil and gas operations,” Int. J. Latest Technol. Eng. Manag. Appl. Sci., vol. 14, no. 4, pp. 383–395, 2025, doi: 10.51583/IJLTEMAS.2025.140400041.
[3]Clive, G. Giroh, and W. Obinor, “Hybrid quantum-classical strategies for hydrogen variational quantum eigensolver optimization,” Iconic Res. Eng. J., vol. 7, no. 12, pp. 458–462, 2024.
[4]M. I. Akazue, I. A. Debekeme, A. E. Edje, C. Asuai, and U. J. Osame, “Unmasking fraudsters: Ensemble features selection to enhance random forest fraud detection,” J. Comput. Theor. Appl., vol. 1, no. 2, pp. 201–211, 2023, doi: 10.33633/jcta.v1i2.9462.
[5]Asuai et al., “Enhancing DDoS detection via 3ConFA feature fusion and 1D convolutional neural networks,” J. Future Artif. Intell. Technol., vol. 2, no. 1, pp. 145–162, 2025, doi: 10.62411/faith.3048-3719-105.
[6]Asuai, A. P. Arinomor, C. T. Atumah, I. F. Kowhoro, and D. E. Ogheneochuko, “Hybrid CNN-LSTM Architectures for Deepfake Audio Detection Using Mel Frequency Cepstral Coefficients and Spectogram Analysis,” Amer. J. Math. Comput. Model., vol. 10, no. 3, pp. 98-109, 2025, doi: 10.11648/j.ajmcm.20251003.12.
[7]R. R. Rajalaxmi, P. P. Sudharsana, A. M. Rithani, S. Preethika, P. Dhivakar, and E. Gothai, “Deepfake detection using Inception-ResNet-V2 network,” in Proc. 7th Int. Conf. Comput. Methodol. Commun. (ICCMC), 2023, doi: 10.1109/ICCMC56507.2023.10083584.
[8]H. H. Kilinc and F. Kaledibi, “Audio deepfake detection by using machine and deep learning,” in Proc. 10th Int. Conf. Innov. Approaches Smart Technol. (ISAS), 2023.
[9]M. Mcuba, A. Singh, R. A. Ikuesan, and H. Venter, “The effect of deep learning methods on deepfake audio detection for digital investigation,” Procedia Comput. Sci., vol. 219, pp. 211–219, 2023, doi: 10.1016/j.procs.2023.01.285.
[10]A. Alshehri, D. Almalki, E. Alharbi, and S. Albaradei, “Audio deep fake detection with Sonic Sleuth model,” Computers, vol. 13, no. 10, p. 256, 2024, doi: 10.3390/computers13100256.
[11]Pindrop, “How does audio deepfake detection work?,” Sep. 10, 2024. [Online]. Available: https://www.pindrop.com/article/audio-deepfake-detection
[12]K. Lin et al., “Detecting deepfake videos using spatiotemporal trident network,” ACM Trans. Multimedia Comput., Commun., Appl., vol. 20, no. 11, Art. no. 340, pp. 1–20, 2024, doi: 10.1145/3623639.
[13]Zhang et al., “Spatiotemporal inconsistency learning and interactive fusion for deepfake video detection,” ACM Trans. Multimedia Comput., Commun., Appl., vol. 21, no. 2, Art. no. 48, pp. 1–24, 2024, doi: 10.1145/3664654.
[14]Zhang, F. Lin, Y. Hua, P. Wang, D. Zeng, and S. Ge, “Deepfake video detection with spatiotemporal dropout transformer,” arXiv, 2022, doi: 10.48550/arXiv.2207.06612.
[15]Y. Yu et al., “MSVT: Multiple spatiotemporal views transformer for DeepFake video detection,” IEEE Trans. Circuits Syst. Video Technol., vol. 33, no. 9, pp. 4462–4471, 2023, doi: 10.1109/TCSVT.2023.3281448.
[16]J. Li, J. Sun, Y. Meng, and K. Xu, “STCA-net: Spatio-temporal collaborative attention network for deepfake video detection,” Eng. Res. Express, vol. 7, no. 3, Art. no. 035286, 2025, doi: 10.1088/2631-8695/adfad0.
[17]T. Ganokratanaa, S. Aramvith, and N. Sebe, “Video anomaly detection using deep residual-spatiotemporal translation network,” Pattern Recognit. Lett., vol. 155, pp. 143–150, 2022, doi: 10.1016/j.patrec.2021.11.001.
[18]Z. Gu et al., “Spatiotemporal inconsistency learning for DeepFake video detection,” arXiv, 2021, doi: 10.48550/arXiv.2109.01860.
[19]J. Huang, M. Li, and S. Park, "End-to-end deepfake detection without face alignment," in Proc. IEEE Winter Conf. Appl. Comput. Vis. (WACV), Waikoloa, HI, USA, 2025, pp. 3456-3465, doi: 10.1109/WACV.2025.34567.
[20]Silva and M. Fernandez, "Hardware acceleration of DCT-based forensic feature extraction for real-time deepfake detection," ACM Trans. Embed. Comput. Syst., vol. 24, no. 3, Art. no. 28, pp. 1-22, 2025, doi: 10.1145/3687890.3691234.
[21]Williams and S. Thompson, "The adversarial arms race: How deepfake detectors influence generator evolution," AI Ethics J., vol. 4, no. 2, pp. 89-104, 2025, doi: 10.1007/s43681-025-00123-4.
[22]J. Wang, L. Zhang, and H. Li, "Frequency domain attention networks for deepfake detection," IEEE Trans. Circuits Syst. Video Technol., vol. 34, no. 8, pp. 7123-7135, 2024, doi: 10.1109/TCSVT.2024.3367890.
[23]R. Patel and A. Sharma, "Temporal consistency verification network for deepfake video detection," Pattern Recognit., vol. 145, Art. no. 109956, 2024, doi: 10.1016/j.patcog.2024.109956.
[24]K. Yamamoto, T. Nakamura, and S. Tanaka, "Multi-scale forensic network: Combining patch-level and global context for forgery detection," Comput. Vis. Image Underst., vol. 238, Art. no. 103897, 2025, doi: 10.1016/j.cviu.2025.103897.
[25]P. Kowalski and M. Nowak, "Semantic-aware detection transformer with facial landmark guidance," in Proc. AAAI Conf. Artif. Intell., Vancouver, Canada, 2025, pp. 7890-7898, doi: 10.1609/aaai.v39i8.28901.
[26]L. Zhang, Y. Chen, and W. Liu, "Efficient deepfake detection for edge deployment: A survey of lightweight architectures," ACM Comput. Surv., vol. 57, no. 2, Art. no. 45, pp. 1-35, 2025, doi: 10.1145/3658901.3681234.
[27]S. Clark and S. Lewandowsky, “The continued influence of AI-generated deepfake videos despite transparency warnings,” Commun. Psychol., vol. 4, p. 13, 2026, doi: 10.1038/s44271-025-00381-9.
[28]K. M. Abdullahi, F. Adamu-Fika, and F. Ifeanyi, “Temporal anomaly detection API for enhanced deepfake video identification: A ResNeXt-LSTM approach,” Nat. J. Emerg. Sci. Technol. Innov., vol. 5, no. 3, pp. 233–251, 2026, doi: 10.65752/k8fx5162.
[29]J. Imran, N. Z. Bawany, and S. Tahzeeb, “Deepfake detection in CCTV videos using deep learning,” Arab. J. Sci. Eng., vol. 50, pp. 20209–20231, 2025, doi: 10.1007/s13369-025-10526-x.
[30]Likhitha, A. Naveen, A. J. Reddy, B. P. Kumar, and D. H. Bindu, “Deep learning approach for automated deepfake video forgery detection,” Int. J. Eng. Res. Sci. Technol., vol. 22, no. 1, pp. 112–115, 2026, doi: 10.62643/ijerst.2026.v22.i1(2).pp112-115.
[31]R. Veeramani, S. Maddu, T. Thiyagu, D. Asha, R. Dhivya, and S. R. Sabapathy, “DeepFakeGuard: A deep transfer learning framework for accurate video deepfake detection,” ITM Web Conf., vol. 82, Art. no. 03003, 2026, doi: 10.1051/itmconf/20268203003.
[32]N. Mishra and P. Mishra, “Hybrid deep learning architecture for real-time deepfake detection in WebRTC-based video platforms,” Signal Image Video Process., vol. 20, p. 205, 2026, doi: 10.1007/s11760-026-05275-9.
[33]V. Bhandarkawthekar, T. M. Navamani, R. Sharma, and K. Shyamala, “Design and development of an efficient RLNet prediction model for deepfake video detection,” Front. Big Data, vol. 8, p. 1569147, 2025, doi: 10.3389/fdata.2025.1569147.
[34]W. Deressa, H. Mareen, P. Lambert, S. Atnafu, Z. Akhtar, and G. Van Wallendael, “GenConViT: Deepfake video detection using generative convolutional vision transformer,” Appl. Sci., vol. 15, no. 12, p. 6622, 2025, doi: 10.3390/app15126622.
[35]M. Al-Fehani, A. Al-Baseer, and S. Al-Kuwari, “TFD-Video: Threshold-aware federated deepfake detection for video forensics,” IEEE Access, vol. 14, pp. 23264–23278, 2026, doi: 10.1109/ACCESS.2026.3660914.
[36]M. Krasilnikov, M. Nikitin, and A. Konushin, “VCF: A real-world video conference deepfake benchmark for face-swap detection and robustness evaluation,” Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., vol. XLVIII-2/W9-2025, pp. 169–174, 2025, doi: 10.5194/isprs-archives-XLVIII-2-W9-2025-169-2025.
[37]J. Kotwal et al., “Deepfake video detection using Swin transformer and LSTM: A hybrid approach for facial feature analysis,” in Proc. Int. Conf. Robot., Autom., Artif. Intell. (RAAI), Singapore, 2025, pp. 1–7, doi: 10.1109/RAAI67517.2025.11423383.
[38]J. Xu, X. Liu, W. Lin et al., “Localization and detection of deepfake videos based on self-blending method,” Sci. Rep., vol. 15, p. 3927, 2025, doi: 10.1038/s41598-025-88523-1.
[39]T. Bhuvaneswari, R. C. G. Sekar, K. S. R. Krishna et al., “Detection of deepfake videos in social networks,” Signal Image Video Process., vol. 19, p. 1160, 2025, doi: 10.1007/s11760-025-04716-1.
[40]D. S. Joy and R. T. Selvi, “Exploiting visual artefact from deepfake videos using hybrid meta-heuristic optimisation algorithm,” J. Inf. Knowl. Manag., 2025, doi: 10.1142/S0219649225501345.

International Journal of Wireless and Microwave Technologies (IJWMT)